>Hi George,
>
>>I'll agree that you probably won't (but might) get better performance with an FLL. My feelings about doing something like that is, "Is the view worth the climb?" Plus, it's just one more thing you have to maintain.
>micro-optimization normally won't win the war if perf is lagging.
>Here it is the case that the routines are used in some of my timing scenarios - whenever I use a vfp based approach
>and try to use high perf timers or measure cpu utilization - and they tend to be called VERY often.
>It is more a case of honing my development environment and vfp "tradition" says you HAVE to spent some time optimizing it. <g>
>
>Regards,
>
thomas,

In general, I don't archictect/design/program for all out performance. I tend to focus on maintainability. In Windows, unless there's an overriding "need for speed", generally, there's little point.

I've been working, however, on some web projects in .NET, and in that environment, I tend to pay a little more attention to the performance factor.

Most recently, I've been working on a project called "Green Metrics". As you might expect, this project tries to measure things like BTUs, Green House Gases, etc. against units produced.

Initally, this was a one-to-one relationship. However, due to "changing requiements" this became a "one-to-the sum of many"

In the beginning, performance was fine. However, after the change in requirements, it was very slow. What was done was that a new SQL Server table was created to store the one-to-one values rather than the calculated one-to-the-sum of many. Now the result is almost instanteous.